Why WAIT doesn't waits exactly for its input?

[Boopatheee]

Hello,

Please find the attachments here.

In Attachment 1 (Wait Program.png) : My actual program was available. The output of the program was unexpected from for the program. The program have an indicator as Waveform Graph to plot the result. The results are attached and explained below.

In Attachment 2 (Output of Waveform Graph.jpg) : I have attached the result of 2 runs which i have performed. Since i have performed the run many times, i noticed this is the big issue for me to continue my project.

I feel that Wait Function was not actually doing its action properly.

This output was not repeated any time again. Every has new functioning was noticed. I attached the image which i have noticed randomly and not repeatedly.

As per program, my output should be constantly 5 (As i given Wait input as 5) and it may be Plus or Minus 1 in very worst case. But i have seen repeatedly it was not exactly the 5, but some times it is 6,7,8,9,..21.  Results attached in Run 2 Graph in attachment 2

The same program i have run it in different Computer (NI Chassis with Controller and Add on Modules), were i have noticed that its around 260 to 272. I have attached the output graph of this in Run 1 Graph in attachment 2.

I have not found anything regarding this issue in ni forum. If this was already discussed and solution available, please share a link of it as a reply.

If you people have any other quires, please post as a reply. Let me clear your doubts.

Thank You

[JÞB]

Well, First your benchmark is calling the ms Timer way too often 😄 Either of these loops would be a better test of the Wait ms timing function.

The Help

offers some insight about potential errors due to the OS.  You CANNOT get deterministic behavior without a RT (Real-Time) OS!

Your PXI chassis sounds like the controller is not receiving correct airflow.  Do you have slot blockers in unused slots? Are the covers in place? Is there adequate clearance around the chassis? Are the Fan Filters clean?

[Boopatheee]

Hi Jay Thornby,

Actually i faced the issue in transferring some data in TCP communication which contains wait inside the loop (had TCP read functions) between Chassis 1 to Chassis 2 (I haven't transfer any big data). Noticed that consuming more time rarely and even it is not affecting my data transfer like missing few data between the extra time delay.

While debugging the extra time delay in the TCP Loop, using the basic test (Attached image of the block diagram earlier) found that it misbehaves. Then tried in different desktops and laptops and noticed that i haven't seen more than 21ms delay in the place of 5ms.

Coming to Chassis Air Flow: 

Chassis kept in a rack with cooling system(Fans on the top of the rack and room consists of Air conditioner, but Air Conditioner will be operating only very few cases) 

Chassis has few unused slots.

All unused slots are not covered with slot blockers.

Ventilation available like chassis was covered on its side and not in the front and back. Even Rack has fans on the top to release the heat.

I'm not sure about the Fan Filters Clean.

I think i have answered your questions. 

Please let me know about it in further to get clear idea about it.

Thanks Mr.Jay Thornby

[JÞB]

---

@Boopatheee wrote:

Hi Jay Thornby,

Actually i faced the issue in transferring some data in TCP communication which contains wait inside the loop (had TCP read functions) between Chassis 1 to Chassis 2 (I haven't transfer any big data). Noticed that consuming more time rarely and even it is not affecting my data transfer like missing few data between the extra time delay.

While debugging the extra time delay in the TCP Loop, using the basic test (Attached image of the block diagram earlier) found that it misbehaves. Then tried in different desktops and laptops and noticed that i haven't seen more than 21ms delay in the place of 5ms.

Coming to Chassis Air Flow: 

Chassis kept in a rack with cooling system(Fans on the top of the rack and room consists of Air conditioner, but Air Conditioner will be operating only very few cases) 

Chassis has few unused slots. All unused slots are not covered with slot blockers. Install the slot blockers Your cooling air is flowing through the unused slots.  Those don't need air to cool them since they don't generate heat get the airflow passing by the controller and any modules that are generating heat!  the first sign of poor cooling is timing drift on the controller due to xtal ovens getting warm.

Ventilation available like chassis was covered on its side and not in the front and back. Even Rack has fans on the top to release the heat. Good- but follow the mounting instructions see pg 2-2 here for your chassis .  You may want to switch the fan speed from "Auto" to "High" see pg 2-3

I'm not sure about the Fan Filters Clean. Same link see pg 3-2 for filter cleaning

I think i have answered your questions. 

Please let me know about it in further to get clear idea about it.

Thanks Mr.Jay Thornby

---

[Boopatheee]

Hi Jay Thornby,

Chassis mounted in the Rack with sufficient spacing as per User manual spacing conditions.

Unused slots was covered with Slot Filler Panel and few left open. Not used any slot blockers to close empty slots.

I'll try by checking the mode of Fan Speed which has to be High instead of Auto

I can try by cleaning the Filter as per procedure given in the user manual.

Is these are the possibilities of time drift ?

Thanks Hi Jay Thornby, 

[JÞB]

---

@Boopatheee wrote:

Hi Jay Thornby,

Chassis mounted in the Rack with sufficient spacing as per User manual spacing conditions.

Unused slots was covered with Slot Filler Panel and few left open. Not used any slot blockers to close empty slots.

I'll try by checking the mode of Fan Speed which has to be High instead of Auto

I can try by cleaning the Filter as per procedure given in the user manual.

Is these are the possibilities of time drift ?

Thanks Hi Jay Thornby, 

---

Yes, these are going to effect timing accuracy of the controller especially if you are running "hot" types of devices next to the controller and have not used slot blockers to force the air to pass by the working equipment.  essentially fans only move a certain volume of air.  if there is "Nothing" in a slot a lot of the air moves through past nothing and nothing gets cooled.  then things that need to be cooled get hotter than you want them to be (there is a reason they put a cooling fan in the system right?)  Most often, the extra heat looks like bad timing

[Boopatheee]

Hi Jay Thornby,

I have few more doubts regarding this issue.

    1. Is any particular temperature will affect Controller performance or it may occur in any temperature?

    2. As i have checked with Controllers User Manual, Operating temperature can be 5 to 50 deg. But i am maintaining the ambient temperature less than 25 deg and controller temperature 32 deg (plus or minus 5 deg).

    3. I have tried the same program (which i have attached on my initial query) in Desktop and few Laptops, even i resulted with 21ms maximum delay in the place of 5ms. This 21ms is not constant every time (may be from 5ms to 21ms). 

   4. If error with 1ms can be adjustable by our customer. But i have noticed that it was not in the limit. RT OS is the only solution to solve this issue?

   5. Are you going to suspect the cooling system of Laptop and Desktop, where i have noticed 21ms as maximum and 5ms as minimum?

Thanks

[JÞB]

Straight to the point.  If you need deterministic timing you must have a RT OS.

Now physics... you have only 1 way to observe "Time"  and that only happens to be "Wait for something to change".  On your PXI chassis controller that is a 32khz crystal oscillator.   Everyone puts them in an oven. Why? Because Temperature changes how fast they can vibrate and keeping them just above ambient temperature is rather easy. So, if cooling air dies not flow around the xtal oven, it gets hotter and faster than intended.

You see? Easy for me to suspect that your original question was a result of "poor cooling " and that you forgot some basic stuff.   No Magic 8-Ball required